Immiscibility between silicate magmas and brines in the Omsukchan Granite
(NE Russia): melt inclusion probe into magmatic-hydrothermal transition
V.S. Kamenetskya,*, V.B. Naumovb, P. Davidsona, E. van Achterberghc, C.G. Ryanc
a
Centre for Ore Deposit Research and School of Earth Sciences, University of
Tasmania, Private Bag 79, Hobart, Tasmania, 7001, Australia
b
Vernadsky Institute of Geochemistry, Russian Academy of Science, Kosigin 19,
Moscow 117975, Russia
c
CSIRO Exploration and Mining, PO Box 136, North Ryde, NSW 2113, Australia
* Corresponding author: Tel.: +613-62267649; fax: +613-62232547
E-mail address: [email protected] (V.S. Kamenetsky).
Abstract
Exsolution (unmixing) of volatile-rich phases (fluids, salt and sulphide melts) from
cooling and crystallising silicate magmas is critical for element transport from the
Earth’s interior into athmosphere, hydrosphere, crustal hydrothermal systems, and the
formation of orthomagmatic ore deposits. Unmixing is an inherently fugitive
phenomenon and melt inclusions (droplets of melt trapped by minerals) provide the
most robust evidence of this process. In this study, melt inclusions in phenocrystic
and miarolitic quartz were studied to better understand immiscibility in the final
stages of cooling of, and volatile exsolution from, granitic magmas, using the tinbearing Omsukchan Granite (NE Russian) as an example.
Primary magmatic inclusions in quartz phenocrysts (cotectic crystals, crystalline
silicate melt, brine- and vapour-rich) demonstrated the coexistence of silicate melt,
salt melt/brine, vapour, and emulsions of these, during crystallisation of the granite
magma, possibly at the waning stage. Microthermometric experiments, in conjunction
with PIXE and other analytical techniques, disclosed extreme heterogeneity in the
composition of the non-silicate phases, even in brine globules within the same silicate
melt inclusion. We suggest that the observed variability is a consequence of strong
chemical heterogeneity in the residual silicate-salt melt system on a local scale, owing
to crystallisation, immiscibility and failure of individual phases to re-equilibrate. The
possible evolution of non-silicate volatile magmatic phases into more typical
“hydrothermal” hypersaline solutions was examined using inclusions in quartz from
associated miarolitic cavities.
Keywords: Melt inclusion; Granite; Immiscibility; Fluid; Brine; PIXE
1
1. Introduction
Volatiles dissolved in aluminosilicate magmas (dominantly H2O, CO2, S and
Cl) play a fundamental role in magmatic evolution across the compositional spectrum
from komatiite to granite, and in the formation of magma related hydrothermal
systems. Volatiles have a major control on rheological properties of magmas,
crystallisation temperature and sequence. When the volatiles are exsolved from
parental magmas, due to the processes of unmixing or immiscibility, and form a
separate volatile-rich phase (fluid), a major change in the element partitioning occurs.
It is believed that volatile phases exsolved during the crystallisation of magmas carry
significant amounts of metallic elements (e.g., Candela and Holland, 1984; Shinohara,
1994; Webster, 1997; Kamenetsky et al., 1999). If dense, these phases may precipitate
in situ (e.g., Fe-Cu-Ni-PGE sulphides); if buoyant, they may enter hydrothermal
systems and be responsible for formation of certain types of ore deposits (e.g., W-Sn
skarns and greisens, pegmatites, and Cu-Mo-Au porphyries).
For many magmas, the evolution of immiscible volatile phases takes place in
the poorly characterised interval between the orthomagmatic and the hydrothermal
regimes of magmatic cooling and crystallisation. This interval near the magmatic
solidus, referred to as transitional with respect to processes and compositions (e.g.,
Burnham and Ohmoto, 1980), has proved extremely difficult to document and
understand, largely because of the transient, reactive nature of volatile phases released
during magmatic cooling. Also, igneous petrology has long been segregated from
studies of hydrothermal systems; although some overlap always existed in the form of
experimental and theoretical studies (e.g., Koster van Groos and Wyllie, 1969;
Candela, 1989; Shinohara et al., 1989; Candela, 1991; Cline and Bodnar, 1991;
Webster, 1992; Shinohara, 1994; Webster, 1997).
In the case of solidifying felsic magmas, volatile phases are largely exsolved
in a temperature window between 850oC and >500oC, which is still remaining
virtually “opaque” to geological research. Apart from the problem of how transitional
volatile phases eventually acquire the properties of the ore-forming solutions, there
are immediate questions about physical state and chemical compositions of late
magmatic fluids. These can be answered by employing melt and fluid inclusion
research on samples representative of magmatic-hydrothermal transition (see reviews
in Roedder, 1984; Roedder, 1992; De Vivo and Frezzotti, 1994; Lowenstern, 1995;
2
Frezzotti, 2001). Despite the contention that “… interpretation of the inclusion record
is commonly ambiguous and difficult, particularly that stage between silicate melt and
the hydrous saline melts and aqueous solutions…” (Roedder, 1992), in recent years
melt and fluid inclusion approach has become more attractive and gained strength
from the use of modern microbeam techniques (e.g., references in Kamenetsky et al.,
2002b). In this presentation we aimed the study of inclusions in phenocrystic and
miarolitic quartz from the Omsukchan granite (NE Russia) to describe an occurrence
of immiscibility between the silicate melt and late magmatic volatile phases,
characterise their composition from microthermometric experiments and in situ
analysis, and attempt to constrain the physical nature and chemical signature of
magmatic derived fluids and their high temperature evolution.
2. Omsukchan granite intrusive
Porphyritic leucogranites of the Cretaceous Omsukchan massif (Chukotka
Peninsula, NE Russia) are located within the Okhotsko-Chukotsky volcanic belt and
host economic tin deposits. The depth of intrusion defined from the thickness of
overlapping volcanisedimentary host rocks was estimated as 0.7-1.5 km (Sokolov,
1980). The intrusion is a flat top cupola with shallow-dipping flanks, containing three
concentric zones of texturally different granites. Mineralisation is localised in the roof
of the cupola, within linear zones of metasomatic Na-Fe alteration and quartztourmaline veins (Sokolov, 1980). Abundant miarolitic cavities lined with large quartz
crystals (to 10 cm) are commonly associated with altered and mineralised granites.
Homogenisation temperatures and salinities of melt and fluid inclusion in
quartz and fluorite from granites, miarolitic cavities and metasomatic ore bodies have
been previously reported (Naumov and Sokolov, 1981; Kamenetsky et al., 2002b).
Temperatures, salinities and compositions were shown to be extremely diverse. To
complement the data on inclusions in miarolitic quartz (Kamenetsky et al., 2002b),
the emphasis of this study is on a suite of inclusions in quartz phenocrysts.
3. Inclusions in quartz phenocrysts
3.1. Occurrence of inclusions
3
Coexisting magmatic inclusions in quartz phenocrysts, unaffected by
hydrothermal alteration, are aligned along growth planes and syn-crystallisation
fractures. There are three main types of inclusions:
(1) Euhedral crystals of feldspar, mica, sphene, magnetite, apatite, zircon etc
(i.e., minerals cotectic with quartz and accidentally trapped);
(2) Ovoid to spherical inclusions of magmatic volatile-rich phase that is
characterised (at room temperature) by variable proportions of aqueous vapour, liquid
and euhedral crystals. The endmembers within this type are represented by essentially
vapour (>90 vol%, Fig. 1a) and essentially crystalline (>70 vol%, Figs. 1b,c)
inclusions. Optical examination revealed at least nine different mineral species (halite
and hematite are tentatively identified) present among solids, and often within a single
inclusion. In highly crystalline volatile-rich inclusions (hereafter, brine or salt melt)
liquid exists interstitially, and a vapour bubble is deformed.
(3) Variable in shape and size birefringent microcrystalline aggregates of
presumably silicate minerals with a misshapen vapour phases filling the spaces
between solids, and rare opaque phases (Figs. 1d,e), interpreted to be inclusions of
crystallised silicate melt. These are commonly surrounded by radiating fractures and
halos of tiny vapour-rich aqueous bubbles, suggesting the melt originally had a high
volatile content, which was released during post-entrapment crystallisation and
decrepitation of these inclusions (Davidson and Kamenetsky, 2001).
As all types of inclusions occur together it is likely that some consist of
heterogeneously trapped combinations of cotectic crystals, silicate melt and aqueous
fluids. An example of such composite inclusion (crystallised silicate melt and halitebearing fluid phase) is shown in Fig. 1f.
3.2. Post-heating occurrence of inclusions
Bulk heating experiments involve heating a quantity of inclusion-bearing
quartz phenocrysts in a 1 atm furnace (850-950oC, 20-50 hours) and consequent air
quenching. One objective was to identify those inclusions that were trapped at
magmatic temperatures and remained uncompromised (unfractured and unleaked).
Bulk heating also revealed inclusions that survived temperature-related volume
change of host quartz (e.g., a-b quartz transition at ~573oC), and thus were suitable
for further microthermometric work. The second aim was to convert crystalline
4
silicate masses within melt inclusions into homogeneous glass. This, in the case of
granitic magmas, usually requires long heating times, and eventually makes
heterogeneously trapped phases (e.g., fluid, if present) available for observation and
analysis. And finally, immiscible brine and fluid phases, if trapped heterogeneously
with silicate melt, can be liberated and further enhanced into larger formations
(globules or bubbles) in order to successfully analyse them using microbeam
techniques.
The post-heating examination of the Omsukchan granite quartz phenocrysts
reveals the presence of small (<10 mm) bubble-free silicate glass inclusions and larger
silicate glass inclusions with one or several low-density vapour bubbles and
sometimes unmelted silicate and opaque phases (Fig. 2a). Commonly, the bubbles are
located in “swellings” on the otherwise rounded surfaces of such inclusions, and this
feature is probably due to differential melting and crystallisation within inclusions
during heating and “sheltering” by the bubble of the inclusion wall against
crystallisation. The majority of silicate glass inclusions contain spherical globules
(Figs. 2c-g) composed of microcrystalline material with variable proportions of
spherical to deformed dark phase (vapour). The number (from 1-2 to 10’s) and sizes
(<1 – 15 mm) of globules vary significantly. Commonly in larger inclusions these
crystal- and bearing-bearing globules form an emulsion (globules are suspended in the
glass “matrix”), and characteristically they are present embedded in the inclusion
walls (Figs. 2c,g and 4b) and often without silicate glass, in the host quartz (Figs.
2d,g, and 3). Rarely, silicate melt inclusions contain a single, relatively small (~3.5
vol%) globule, as in the two coexisting inclusions shown in Fig. 2e. Other inclusions
show a globule (similar to those described above) rimmed by relatively small volume
of the silicate glass (Fig. 2b). And finally, inclusions that contained brine, pre-heating,
are present as negative-crystal shaped inclusions of crystalline material and deformed
bearing bubble (Figs. 2h,i).
3.3. Heating stage experiments with melt and brine inclusions
Heating stage experiments were performed to observe, document (e.g.,
photographing or video recording) and register temperature of phase transformations
(such as melting, homogenisation, crystallisation etc) inside individual inclusions. A
5
Linkam TS1500 heating stage with TMS 94 temperature control unit and an Olympus
BX51 microscope with a DP11 digital camera were used. Heating/cooling rates (from
5 to 100oC/min) were varied when required for more efficient documentation. All
grains were first bulk heated in the furnace and quenched (see above). Heating stage
studies targeted spherical globules inside clear silicate glass, and also those trapped in
quartz with no silicate glass around them.
Examples of experiments with spherical globules present in the silicate glass
are given in Fig. 4. Several subsequent heating runs for each inclusion showed
consistent results. Although, the behaviour of all globules studied was very similar,
the temperatures of phase transformations in different samples varied. The following
general observations were made for spherical globules during heating (Fig. 4):
(1) They remain constant in size, do not change their position with respect to
silicate phase (melt), and do not mix with silicate melt even at temperatures exceeding
liquidus (900-1100oC);
(2) The first visible changes occur at 120-190oC. These include movement and
re-arrangement of solids and vapour bubble, enhancement of phase boundaries, and
the appearance of new phases;
(3) The vapour phase acquires spherical shape at 250-350oC, and the
boundaries between solid phases become invisible;
(4) In some inclusions vapour bubbles disappear at 420-510oC, and then
renucleate at ~580oC. In other globules there is no large change in the bubble size
until the quartz transition temperature (~573oC), in some cases bubbles increase
significantly, up to almost the size of the globule;
(5) Melting of solid phase(s) becomes noticeable at >600oC and is seen as
appearance of liquid (melt) surrounding a cubic crystal;
(6) Heating above 630oC results in rapid decrease of the bubble and crystal(s),
with complete melting occurring at 670-735oC. At these temperatures the liquid in
globules has lower refractive index and low viscosity than silicate melt;
(7) Vapour bubbles shrink and move randomly and rapidly, until final
disappearance at 790-935oC;
(8) At the temperature of bubble homogenisation an elongate prismatic
brownish crystal, possibly formed as a result of reaction between salt and silicate
melt, may remain on the globule-silicate melt interface (Fig. 4a).
6
It is worth noting that, irrespective of their size, the above phase
transformations take place almost simultaneously in all entrapped globules within an
individual silicate melt inclusion. This may suggest similar compositions and
entrapment conditions. The observed differences between globules in different
inclusions can be tentatively attributed to the effects of internal pressure.
A record of typical heating experiment with a glass-free, spherical inclusion
embedded in quartz is shown in Fig. 5. In this case, a clear cubic crystal melted at
630oC, and the melting of a brownish elongate crystal and homogenisation into liquid
occurred at 890oC. The melting and homogenisation temperatures can vary by 100150oC for several tens inclusions studied, even for those within same quartz
phenocryst.
4. Inclusions in miarolitic quartz
Miarolitic cavities within the Omsukchan granite are filled with euhedral
quartz crystals that contain numerous populations of inclusions, mainly aligned along
syn-crystallisation fractures and more rarely along the growth planes. The majority of
inclusions are crystallised saline aqueous solutions (brines), with some vapour-rich
inclusions. Silicate melt inclusions have not been recorded. The earlier studies of
brine inclusions in the Omsukchan miarolitic quartz (Naumov and Sokolov, 1981;
Kamenetsky et al., 2002b) recorded their large size (to 300 mm), high homogenisation
temperatures (to 840oC), and diverse and often metal-enriched chemical
compositions. We emphasise the significance of miarolitic brine inclusions for this
study is in that they probably represent a magmatic volatile phase that separated from
the residual melt at the brink of its solidification. Therefore, they can compliment the
results for the fluids exsolved earlier, and trapped as brine globules in phenocrystic
quartz. Moreover, the exceptionally large size of miarolitic brine inclusions makes it
possible to better document assemblage of daughter minerals, and phase
transformation during heating.
4.1. Inclusion paragenesis
The fact that multiphase brine inclusions belong to the crosscutting trapping
planes, and thus to the different trapping events (Roedder, 1979), calls for
7
development of criteria for recognition of “assemblages” or “texturally coeval”
groups of inclusions. In the following two examples we present descriptions and
measurements of phase proportions for inclusions that coexist within a section of the
trapping plane (between cross-cutting fractures, Figs. 6a,d), and thus are considered to
be trapped simultaneously as a homogeneous solution.
The first example concerns a simple case of the four-phase (liquid + vapour
bubble + cubic crystal (halite?) + opaque crystal) inclusions. Despite significantly
variable size (7-180 mm) thirty-two measured inclusions in this plane (Figs. 6a-c)
have nearly constant volumes of vapour bubble and cubic mineral relative to the
volume of inclusion (0.127±10% and 0.043±15%, respectively). In the second
example (Figs. 6d-f) thirty-four multiphase (liquid + vapour bubble + 3 to 6 crystals)
inclusions of different size (13-120 mm) show similar and constant bubble/inclusion
volume ratios (0.127±12%) that in this case seem to be characteristic of homogeneous
trapping of aqueous saline fluids. Daughter minerals are typically euhedral, although
crystalline aggregates are also present, and show a variety of crystal forms and optical
properties, including colour, relief and birefringence (Kamenetsky et al., 2002b).
Different appearance of the assemblages of daughter crystals as well as their different
compositions (confirmed by laser Raman spectroscopy, Kamenetsky et al., 2002b), is
likely to reflect divergent post-entrapment crystallisation paths for individual
inclusions. Moreover, even at room temperature the content of multiphase inclusions
is metastable and prone to spontaneous recrystallisation followed by changes in a
number, shape and volume ratios of phases (Fig. 7, see also Lowenstern, 1994). This
makes it problematic to distinguish between multiphase inclusions belonging to
different trapping events.
4.2. Heating stage experiments with multiphase inclusions
Due to their small size, the behaviour of spherical brine globules in quartz
phenocrysts during heating is not fully discernible (Fig. 5). However, as the formation
of miarolitic quartz immediately postdated quartz phenocrysts, the microthermometry
of larger miarolitic multiphase inclusions can be cautiously used to characterise
earlier granitic brines.
8
As expected from the variable appearance of inclusions at room temperature
(e.g., Figs. 6e,f), the melting behaviour even in those that seemingly belong to the
same trapping plane, is also distinct. The summary of general features is as follows
(Fig. 8). Most crystals in the multiphase inclusions dissolve below 300-350oC. The
first crystal usually melts at 55-60oC, the second – at 130-150oC, the third and fourth
disappear at 250-300oC. The large isotropic phase with low relief and cubic shape
(presumably halite) melts at 300-625oC (554oC in the example in Fig. 8a). Breakdown
of some birefringent phases at T~ 400oC results in the formation of dispersed dark
particles (photo at 390oC in Fig. 8c) that coalesce into semi-transparent aggregate of
high relief prismatic crystals (photo at 554oC in Fig. 8a and photo at 398oC in Fig.
8c). These newly formed phase(s) starts to melt at 620-630oC, but usually persists to
much higher temperatures (~800oC). Bubbles dissolve at variable temperatures from
380 to 840oC. In the example shown (Fig. 8a) a bubble first disappeared at 480oC,
then re-appeared at 593oC (just above the temperature of quartz transition), and
disappeared again at 618oC. Relatively large inclusions (>50 mm) often decrepitate at
temperatures below complete homogenisation (photo at 650oC in Fig. 8b; note that an
aggregate of crystals is still present) with the inclusion content expelled at the grain
surface through a fracture. Quenching of the inclusions results in formation of
aggregates of prismatic crystals at <250oC that almost fill the inclusion (Fig. 8d),
whereas slow cooling produces a large cubic crystal (halite?) at <420oC, followed by
spontaneous crystallisation of prismatic birefringent crystals at <200oC (Fig. 8e).
5. Composition of globules and brine inclusions
Low temperatures of the first and final melting in the globular inclusions in
quartz phenocrysts and multiphase inclusions in miarolitic quartz, the low viscosity of
their content inferred from heating runs (Figs. 4, 5 and 8), and the presence of
aqueous bearing and liquid are suggestive of essentially non-silicate, volatile-rich
compositions for these inclusions. Furthermore, the presence of cubic daughter
crystals indicates a significant chloride component. Due to their low viscosity
chloride-rich aqueous fluids and melts (as well as other non-silicate melts) are not
quenchable into homogeneous phase (glass). Consequently, at room temperatures the
studied inclusions contain aggregates of crystals, a large volume vapour bubble(s) and
interstitial aqueous solution (e.g., Figs. 2h,i and 8d,e). There are many difficulties in
9
calculating bulk inclusion composition based on the compositions of individual
phases and their mass proportions, most immediately, loss of aqueous and gaseous
components when such inclusions are exposed. Additionally, the solid content of
inclusions is hygroscopic. The analytical methods applied in this case (scanning
electron and nuclear microscopy) were partly successful in detecting some elements
present and their abundances, although quantitative analysis of fluid and brine
inclusions, especially their volatile components, is still under development.
Commonly, inclusions close to the surface of host grains decrepitate during
heating and emptied their contents on to the surface (Figs. 8b and 9). We used a
scanning electron microscope (Electroscan ESEM2020 equipped with a Link Pentafet
SATW energy dispersive X-ray detector, University of Tasmania) to estimate the
composition of crystalline and amorphous phases deposited on the surface of host
quartz as a result of evaporation of the ejecta from decrepitated (bursting) non-silicate
inclusions (Fig. 9). Although the ejecta is heterogeneous in phase and chemical
composition, it is typically characterised by the dominance of Na, S, Fe, Ca, and Mn.
Potassium and Zn are minor components, whereas Cl and Cu are completely missing
from the X-ray spectra (Fig. 9). Because some elements are more volatile than others
(Cl and Cu in this case) the quantification of the bulk inclusion composition using this
method is not possible.
The distribution and abundance of elements heavier than Cl within unopened
inclusions were studied by the proton-induced X-ray emission method (PIXE) using
the CSIRO-GEMOC Nuclear Microprobe (e.g., Ryan et al., 2001a; Ryan et al.,
2001b). The PIXE data on the multiphase inclusions in miarolitic quartz, reported
earlier (Kamenetsky et al., 2002b), have demonstrated significant concentrations of Cl
(up to 50 wt%), K, Fe and Pb (up to 20 wt%), Ca (up to 6 wt%), Mn and Zn (up to 3
wt%), and Cu, As, Br, Rb, Sr and Sn (up to 1000-3000 ppm). The concentrations of
individual elements were 20-90 times higher in the high-salinity (50 wt% Cl) crystalrich inclusions compared to the low-salinity (5 wt% Cl) liquid-rich type. Given the
highly variable element ratios in these inclusions a conclusion was made that the
granite-derived aqueous saline fluids (brines), if represented by quartz-hosted
inclusions, were extremely heterogeneous in composition at the time of quartz growth
(Kamenetsky et al., 2002b). These results are complemented in this study by the PIXE
element images and concentrations obtained from the spherical brine globules in
quartz phenocrysts (Fig. 10 and Table 1). It appears that both types of spherical
10
globules (i.e. trapped with silicate melt and included in quartz) have the same
geochemical signature. They are enriched in Cl (and probably Na, not analysed by
PIXE), K, Fe, Cu, and to a lesser extent in Ca, Mn, Zn and Pb. Even elements with
concentrations of several hundred ppm (Se, As, and Br) are regarded here as highly
enriched relative to average crustal abundance, and granitic magmas, in particular.
The partitioning of a number of analysed elements between co-trapped silicate
melt and globules is shown in Fig. 10. The intensity of secondary X-rays for a given
element depends on the concentration, density, and depth from the surface. For
example, in Fig. 10a the signal from Cl (a relatively light element with low energy Xrays) was strongly attenuated for a deep-seated globule, whereas the intensities of a
Cu signal for this, and the shallow globule are comparable. Attenuation of low energy
X-rays and uncertainties in the inclusion geometry, depth and density increase the
error in the calculated concentrations. However, these uncertainties can be reduced if
the ratios of elements of similar mass (and with similar X-ray energy), such as Fe and
Mn, Cu and Zn, As and Se, etc are used. The reliability of element ratios, analysed by
PIXE, is independently confirmed by the laser ablation ICPMS analyses of fluid
inclusions (L. Danyushevsky, pers. comm.).
A number of the element ratios for two globules enclosed within a single
silicate melt inclusion (Fig. 10a) are noticeably different (Table 1). Although there
exists broad chemical similarity for pairs of spatially associated brine globules in the
same quartz phenocryst (Fig. 10c, Table 1, sample DK1-8, pairs of inclusions B - C,
and D – E as shown in Fig. 3), the average element ratios for each pair of inclusions
are significantly different. Furthermore, the comparison between all analysed brine
globules in quartz phenocrysts show considerable variability (>25%) in many element
ratios (Table 1).
Tin was detected in brine inclusions in miarolitic quartz in relatively high
amounts (up to 1700 ppm). Tin resides in minute crystalline phases that are possibly
not cassiterite, as other elements (Cl ± As ± Mn ± Ca) are also present (Fig. 11 and
figure 3D from Kamenetsky et al., 2002b).
6. Discussion and conclusions
6.1. Evidence for immiscibility
11
Since the pioneering work on the immiscibility between silicate melts and
dense saline fluids in the Ascension Island granites (Roedder and Coombs, 1967),
only a few studies of intrusive samples have elaborated on the problem of exsolution
of brines/ salt melts from, and coexistence with, deep-seated felsic magmas.
Specifically, the unambiguous records of the silicate and salt melts co-trapping in a
single composite inclusions provided by the studies of the Mariktikan granite
intrusion, Buryatia, Russia (Reyf and Bazheyev, 1977), Monte Genis leucogranite,
Sardinia, Italy (Frezzotti, 1992) and felsic xenoliths from Ventotene Island, Italy (De
Vivo et al., 1995). The reasons for apparent lack of more widespread evidence for the
silicate melt – brine (salt melt) immiscibility have been previously discussed
(Roedder, 1984; Roedder, 1992), and we concur that the presence of dispersed,
essentially non-silicate phases in the silicate magma is, to certain extent,
disadvantageous for the trapping of silicate melt inclusions. However, we consider the
“human factor” more important, for many years these inclusions have not been
recognised in quartz from intrusive rocks and “typically hydrothermal” quartz veins.
Given the results of this study, we suggest that the application of heating and
quenching to several hundred quartz grains should be routinely used to identify the
presence of the silicate glass (with or without shrinkage bubbles and heterogeneously
trapped volatile-rich phases), otherwise obscured by post-entrapment processes. In the
case of quartz phenocrysts from the Omsukchan granite this technique revealed a
clear case of immiscibility with at least two liquids simultaneously present during
quartz growth, with a meniscus between them. This is best demonstrated by the
emulsion textures of many inclusions, in which multiple brine globules (i.e., the
dispersed phase) are randomly distributed within silicate glass (Figs. 2c-g) that at the
time of trapping was a continuous melt phase from which quartz phenocrysts
crystallised. The size of the trapped continuous melt phase is random, largely
depending on the size of the growth irregularities on the surface of the host mineral
(Roedder, 1984), whereas the size of the trapped dispersed phase is representative of
the actual size of emulsion (Reyf, 1984). In other words, the size of the salt melt
globules present in the silicate melt inclusions (<1 – 15 mm; Figs. 2c-g) is
characteristic of the size of natural dispersed phase at the time of immiscibility and
before the coalescence.
Although heterogeneous trapping of immiscible volatile-rich (brine) and
volatile-poor (silicate melt) phases, as shown by the emulsion textures, is most
12
common in the case of the Omsukchan melt inclusions, some inclusions may still
represent homogeneous trapping of either silicate or salt melt “endmembers” (Figs.
2a,h,i and 3). As the number of brine inclusions with no traces of the continuous
phase exceeds the number of homogeneous silicate glass inclusions by several orders
of magnitude, we conclude that the trapping of the discontinuous brine phase is
preferred, possibly because of better wetting properties (Roedder, 1984). Endmember
phases present in variable proportions within a single two-phase inclusions (Figs. 2bg) further advocate heterogeneous trapping, especially when the amount of brine is
relatively large (>20%). In cases with only a small proportion of brine (Fig. 2e),
heterogeneous trapping is less certain because in situ immiscibility is at least possible.
The temperatures of phase transformations and homogenisation within quartzhosted inclusions recorded in this study (Figs. 4, 5 and 8; see also Naumov and
Sokolov, 1981), although important in generally supporting the presence of essentially
non-silicate components in the granitic magma at >650-700oC, are not applicable to
direct quantification of crystallisation temperatures. This is due largely to several
unknown parameters controlling internal pressure, and thus experimental
temperatures. The most important are the compressibility of silicate and salt melts
(e.g., Roedder and Bodnar, 1980) and their effect on each other in the case of
heterogeneously trapped inclusions, the effects of a-b quartz transition and related
microfracturing.
6.2. Compositions of immiscible non-silicate phases
Based on the phase composition of inclusions in quartz phenocrysts in the
Omsukchan granite, the crystallising system consisted of at least silicate melt, saline
fluid (brine) and low-density vapour (Figs. 1-3). We interpreted these phases as
formed due to immiscibility of the precursor granitic magma that happened before or
during quartz crystallisation. As in many other studies our data is not sufficient to
resolve whether there was an immiscible fluid of moderate salinity from which vapour
and brine formed (Burnham, 1979), or whether vapour and brine were exsolved
directly from the magma (e.g., Cline and Bodnar, 1994; Shinohara, 1994; Bodnar,
1995). What remains certain is that two non-silicate, volatile-rich phases (i.e., brine
and vapour) have been coexisting with each other and with the silicate melt, possibly
over significant temperature interval. This provides the most compelling support for
13
the idea that “…granite is only an intermediate stage on a long evolutionary line of
melt to ‘fluid’ compositions. Crystallisation of volatile-bearing granitic magmas can
result in a wide range of fluid compositions, at various P, T, and degrees of
crystallisation” (Roedder, 1984).
In our study a wide range of “fluid” compositions, as sampled by brine
globules in phenocrystic quartz, is also recorded, despite a limited number of analyses
(Table 1). The globules are all metal-rich chloride complexes (Fig. 10, Table 1),
although the element abundances and ratios vary significantly, even for co-trapped
brine globules (Figs. 4a and 10a). There may be numerous reasons for this, e.g.,
exsolution of globules (now co-trapped) from different magma batches. But the most
immediate is an extreme fractionation of most elements between immiscible silicate
melt and non-silicate “fluids” and the disequilibrium character of exsolution. The
latter means that if immiscibility is a continuous process, the components of dispersed
phase must have varying composition because of relatively slow diffusion rates. Here
we regard disequilibrium as a scale-dependent phenomenon, especially in the residual
granitic system, where crystallisation and immiscibility drive chemical fractionation
to the extreme. In practice, two adjacent immiscible brine globules formed
consecutively, sample melts of significantly different compositions due to progressive
depletion in elements compatible with a brine, by the earlier nucleated globule. But as
diffusion rates vary for different elements, even the element ratios in adjacent,
consecutively formed globules may be expected to fluctuate (Fig. 10a, Table 1).
Although it is expected that coalescence of immiscible volatile-rich phases
may reduce chemical variability, however, we anticipate that different batches of
magmatic “fluids” have different histories in transit to miarolitic cavities (in effect
fluid chambers) where they accumulate. If this is the case, discrete batches of fluids,
following a rock-dominated chemical path (e.g., Heinrich, 1990), are prone to
modifications caused by interaction with a granite (solid rocks and residual magma
pockets). Moreover, implied unmixing of the fluids due to boiling or effervescence,
cooling and related crystallisation, and mixing of fluid batches with each other and
possibly external fluids can be responsible for even greater fluid variability, and the
change of primary magmatic geochemical signature, as observed in inclusions trapped
by miarolitic quartz (Table 1; see also Kamenetsky et al., 2002b). Although some
element ratios remain relatively constant for brine inclusions in earlier phenocrystic
quartz, and later miarolitic quartz, the concentration of Cu and proportion of Cu to
14
other metals decrease dramatically in miarolitic brines (from wt% to hundreds of
ppm, Table 1; Kamenetsky et al., 2002b). Such discontinuous evolution of the fluid
and the absence of Cu mineralisation in the Omsukchan granite pose the question
whether magmatic fluids in passage to miarolitic cavities experienced separation and
loss of vapour-rich phase in which Cu strongly partitions (e.g., Lowenstern et al.,
1991; Heinrich et al., 1992; Lowenstern, 1993; Heinrich et al., 1999; Kamenetsky et
al., 2002a)? Other questions also arise, one is the fate of other economic metals (e.g.,
Pb and Zn) that are present in significant amount in brine inclusions but did not form
mineralisation. Similarly, despite strong influx of iron during metasomatism of the
Omsukchan granite (Sokolov, 1980), Fe sulphides are virtually absent, implying the
lack of sulphur in the system.
In conclusion we emphasise that “…formation and evolution of the many
different types of immiscible fluids is complex …but (interpretations) may be of great
value in understanding magmatic processes and the resultant ore deposits.” (Roedder,
1992), and thus the research into evolution of truly magmatic fluids to hydrothermal
solutions is timely important and will require an international multidisciplinary effort.
Acknowledgements
We are grateful to David Steele for help with the electron microscopy. This paper has
benefited from discussions with R. Bodnar, C. Heinrich, J. Lowenstern, T.
Sushchevskaya, and J. Webster. The research was supported by the Australian
Research Fellowship and the Centre for Ore Deposit Research (University of
Tasmania) grants to V.S. Kamenetsky and Russian Foundation for Basic Research
grant 01-05-64109 to V.B. Naumov.
15
References
Bodnar, R.J., 1995. Fluid-inclusion evidence for a magmatic source of metals in
porphyry copper deposits. In: Thompson, J.F.H. (Ed.), Magmas, fluids, and
ore deposits. Miner. Assoc. Canada Short Course Ser., pp. 139-152.
Burnham, C.W., 1979. Magmas and hydrothermal fluids. In: Barnes, H.L. (Ed.),
Geochemistry of Hydrothermal Ore Deposits. J.Wiley & Sons, New York, pp.
71-136.
Burnham, C.W. and Ohmoto, H., 1980. Late-stage processes of felsic magmatism.
Mining Geol. Spec. Iss. 8, 1-11.
Candela, P.A., 1989. Calculation of magmatic fluid contributions to porphyry-type
ore systems: Predicting fluid inclusion chemistries. Geochem. J. 23, 295-305.
Candela, P.A., 1991. Physics of aqueous phase evolution in plutonic environments.
Am. Mineral. 76, 1081-1091.
Candela, P.A. and Holland, H.D., 1984. The partitioning of copper and molybdenum
between silicate melts and aqueous fluids. Geochim. Cosmochim. Acta 48,
373-380.
Cline, J.S. and Bodnar, R.J., 1991. Can economic porphyry copper mineralization be
generated by a typical calc-alkaline melt? J. Geophys. Res. 96, 8113-8126.
Cline, J.S. and Bodnar, R.J., 1994. Direct evolution of brine from a crystallizing
silicic melt at the Questa, New Mexico, molybdenum deposit. Econ. Geol. 89,
1780-1802.
Davidson, P. and Kamenetsky, V.S., 2001. Immiscibility and continuous melt-fluid
evolution within the Rio Blanco porphyry system, Chile: Evidence from
inclusions in magmatic quartz. Econ. Geol. 96, 1921-1929.
De Vivo, B. and Frezzotti, M.L., 1994. Evidence for magmatic immiscibility in Italian
subvolcanic systems. In: de Vivo, B. and Frezzotti, M.L. (Eds.), Fluid
inclusions in minerals: methods and applications. Short course of the working
group IMA. Virginia Tech., Pontignano-Siena, pp. 345-362.
De Vivo, B., Torok, K., Ayuso, R.A., Lima, A. and Lirer, L., 1995. Fluid inclusion
evidence for magmatic silicate/saline/CO2 immiscibility and geochemistry of
alkaline xenoliths from Ventotene Island, Italy. Geochim. Cosmochim. Acta
59, 2941-2953.
Frezzotti, M.L., 1992. Magmatic immiscibility and fluid phase evolution in the Mount
Genis granite (southeastern Sardinia, Italy). Geochim. Cosmochim. Acta 56,
21-33.
Frezzotti, M.L., 2001. Silicate-melt inclusions in magmatic rocks: applications to
petrology. Lithos 55, 273-299.
Heinrich, C.A., 1990. The chemistry of hydrothermal tin (-tungsten) ore deposition.
Econ. Geol. 85, 457-481.
Heinrich, C.A., Günther, D., Audétat, A., Ulrich, T. and Frischknecht, R., 1999. Metal
fractionation between magmatic brine and vapor, determined by microanalysis
of fluid inclusions. Geology 27, 755-758.
Heinrich, C.A., Ryan, C.G., Mernagh, T.P. and Eadington, P.J., 1992. Segregation of
ore metals between magmatic brine and vapor: a fluid inclusion study using
PIXE microanalysis. Econ. Geol. 87, 1566-1583.
Kamenetsky, V.S., Wolfe, R.C., Eggins, S.M., Mernagh, T.P. and Bastrakov, E.,
1999. Volatile exsolution at the Dinkidi Cu-Au porphyry deposit, Philippines:
A melt-inclusion record of the initial ore-forming process. Geology 27, 691694.
16
Kamenetsky, V.S., Davidson, P., Mernagh, T.P., Crawford, A.J., Gemmell, J.B.,
Portnyagin, M.V. and Shinjo, R., 2002a. Fluid bubbles in melt inclusions and
pillow-rim glasses: high-temperature precursors to hydrothermal fluids?
Chem. Geol. 183, 349-364.
Kamenetsky, V.S., van Achterbergh, E., Ryan, C.G., Naumov, V.B., Mernagh, T.P.
and Davidson, P., 2002b. Extreme chemical heterogeneity of granite-derived
hydrothermal fluids: An example from inclusions in a single crystal of
miarolitic quartz. Geology 30, 459-462.
Koster van Groos, A.F. and Wyllie, P.J., 1969. Melting relationships in the system
NaAlSi3O8 -NaCl -H2O at one kilobar pressure, with petrological applications.
J. Geol. 77, 581-605.
Lowenstern, J.B., 1993. Evidence for a copper-bearing fluid in magma erupted at the
Valley of Ten Thousand Smokes, Alaska. Contrib. Mineral. Petrol. 114, 409421.
Lowenstern, J.B., 1994. Chlorine, fluid immiscibility, and degassing in peralkaline
magmas from Pantelleria, Italy. Am. Mineral. 79, 353-369.
Lowenstern, J.B., 1995. Applications of silicate-melt inclusions to the study of
magmatic volatiles. In: Thompson, J.F.H. (Ed.), Magmas, fluids, and ore
deposits. Miner. Assoc. Canada Short Course Ser., pp. 71-99.
Lowenstern, J.B., Mahood, G.A., Rivers, M.L. and Sutton, S.R., 1991. Evidence for
extreme partitioning of copper into a magmatic vapor phase. Science 252,
1405-1409.
Naumov, V.B. and Sokolov, A.L., 1981. Genetic correlation of granites and the
Industrialnoe tin ore deposit according to fluid inclusions data. Geol. Ore
Depos. 4, 74-80.
Reyf, F.G., 1984. Microemulsion state of fluid saturated granitic magmas:
characteristics and petrological implications. Trans. (Doklady) USSR Acad.
Sci. 276, 1197-1201.
Reyf, F.G. and Bazheyev, Y.D., 1977. Magmatic chloride solutions and tungsten
mineralizations. Geochem. Inter. 14, 45-51.
Roedder, E., 1979. Fluid inclusions as samples of ore fluids. In: Barnes, H.L. (Ed.),
Geochemistry of Hydrothermal Ore Deposits. Wiley & Sons, New York, pp.
684-737.
Roedder, E., 1984. Fluid inclusions. Reviews in mineralogy, Vol. 12. Book Crafters
Michigan.
Roedder, E., 1992. Fluid inclusion evidence for immiscibility in magmatic
differentiation. Geochim. Cosmochim. Acta 56, 5-20.
Roedder, E. and Bodnar, R.J., 1980. Geologic pressure determinations from fluid
inclusion studies. Ann. Rev. Earth Planet. Sci. 8, 263-301.
Roedder, E. and Coombs, D.S., 1967. Immiscibility in granitic melts, indicated by
fluid inclusions in ejected granitic blocks of Ascension Island. J. Petrol. 8,
417-451.
Ryan, C.G., McInnes, B.M., Williams, P.J., Dong, G., Win, T.T. and Yeats, C.J.,
2001a. Imaging fluid inclusion content using the new CSIRO-GEMOC
Nuclear Microprobe. Nucl. Instr. & Methods in Physics Res. Sect. B-Beam
Interactions with Materials & Atoms 181, 570-577.
Ryan, C.G., van Achterbergh, E., Griffin, W.L., Pearson, N.J., O'Reilly, S.Y. and
Kivi, K., 2001b. Nuclear microprobe analysis of melt inclusions in minerals:
Windows on metasomatic processes in the Earth's mantle. Nucl. Instr. &
17
Methods in Physics Res. Sect. B-Beam Interactions with Materials & Atoms
181, 578-585.
Shinohara, H., 1994. Exsolution of immiscible vapor and liquid phases from a
crystallizing silicate melt: Implications for chlorine and metal transport.
Geochim. Cosmochim. Acta 58, 5215-5221.
Shinohara, H., Iiyama, J.T. and Matsuo, S., 1989. Partition of chlorine compounds
between silicate melt and hydrothermal solutions. I. Partition of NaCl-KCl.
Geochim. Cosmochim. Acta 53, 2617-2630.
Sokolov, A.L., 1980. Metasomatic zonation of the Industrialnoe-Khatarenskoe ore
field (Omsukchan district). Geol. Ore Depos. 2, 27-35.
Webster, J.D., 1992. Water solubility and chlorine partitioning in Cl-rich granitic
systems: Effects of melt composition at 2 kbar and 800oC. Geochim.
Cosmochim. Acta 56, 679-687.
Webster, J.D., 1997. Exsolution of magmatic volatile phases from Cl-enriched
mineralizing granitic magmas and implications for ore metal transport.
Geochim. Cosmochim. Acta 61, 1017-1029.
18
Figure captions
Fig. 1. Magmatic inclusions in quartz phenocrysts. (a) coexisting vapour-rich (dark
bubbles) and multiphase brine inclusions; (b, c) multiphase, crystal-rich brine inclusions;
(d-f) silicate melt inclusions consisting of birefringent material of different crystallinity,
opaques, vapour phases (dark) and sometimes non-silicate minerals. Note cubic crystal,
possibly halite, in the lower right corner of the inclusion in (f). Scale bars are 20 mm.
Fig. 2. Magmatic inclusions in quartz phenocrysts after bulk heating and quenching (see
text for details), except in (g) photographed at ~850oC . (a) silicate glass with shrinkage
bubble; (b) composite inclusions consisting of a spherical brine globule and a relatively
small amount of silicate glass; (c-g) variable proportions of coexisting brine globules and
silicate glass in individual inclusions. Note emulsion texture in (c, f, and g), and the
presence of similar brine globules in the host quartz around these inclusions in (d and g).
(h, i) microcrystalline structure and deformed vapour phase (dark) of rapidly quenched
brine inclusions. Scale bars are 10 mm. Compositions of globules in coexisting inclusions
in (e) are in the Table 1 (# 3 and #4); PIXE element maps for DK1-7A are shown in Fig.
10b.
Fig. 3. Spherical and ovoid inclusions of brine in a single quartz phenocryst (positions are
circled). Inclusion size is ~10-12 mm. Compositions of inclusions are in the Table 1 (# 58); PIXE element maps for DK1-8E are shown in Fig. 10c.
Fig. 4. Phase transformations in spherical brine globules, co-trapped with silicate melt,
during heating stage experiments. Scale bars are 15 mm. Compositions of brine globules
in the inclusion in (a) are in the Table 1 (# 1 and #2); their PIXE element maps are shown
in Fig. 10a.
Fig. 5. Phase transformations in the spherical brine globule, entrapped in quartz
phenocryst (no silicate glass present), during a heating stage experiment. Scale bar is 15
mm.
Fig. 6. Multiphase inclusions of aqueous saline fluid within trapping planes in miarolitic
quartz. Crosscutting planes are shown by arrows in (a, d).
(b, c) and (e, f) representative coexisting inclusions from the trapping planes shown in (a)
and (c), respectively (in labelled circles). Similar inclusions were reported by
Kamenetsky et al., 2002.
Fig. 7. Metastable nature of multiphase brine inclusions in miarolitic quartz, shown by
spontaneous recrystallisation of daughter crystals at room temperature. Photographs of
inclusions are taken: (b) 270 days after (a), and (d) 188 days after (c). Similar inclusions
were reported by Kamenetsky et al., 2002.
Fig. 8. Phase transformations in multiphase brine globules during heating stage
experiments in miarolitic quartz,. (b) decrepitated inclusion shown in (a). (d, e) Rapidly
(~100ºC/min) and slowly (5ºC/min) cooled inclusion shown in (c). Inclusion sizes are
~60 mm. Similar inclusions were reported by Kamenetsky et al., 2002.
Fig. 9. Scanning electron photomicrographs and representative EDS spectrum of
polymineral aggregates crystallised on the surface of miarolitic quartz, from ejected
contents of decrepitated during experimental heating brine inclusions. Elevated
abundances of Si in the spectrum are from host quartz.
Fig. 10. Optical images and PIXE element maps of heated inclusions of silicate glass and
brine (a, b) and unheated brine inclusion (c) in quartz phenocrysts. Mapping was
performed using a 0.3–0.7 nA beam of 3 MeV protons focused into an ~2 mm beam spot
with a CSIRO-GEMOC nuclear microprobe (Ryan et al., 2001a, 2001b). Legend shown
in lower-right image of (c) indicates relative intensity (and, thus, concentration). Colour
scale in each element image is normalised to its own maximum. Outlines on element
maps mark boundaries of inclusions and brine globules. Corresponding compositions are
given in Table 1: #1 and #2 for shallow and deeper globules, respectively upper and
lower globules in (a); #3 for a globule in (b) and #8 for inclusion in (c).
Fig. 11. Scanning electron photomicrograph of a broken surface of miarolitic quartz and
an EDS spectrum of the crystal in the cavity of the exposed (presumably) brine inclusion.
Table 1. Compositions and selected element ratios of individual spherical brine
globules in quartz phenocrysts, and average composition of multiphase saline fluid
inclusions in miarolitic quartz, analysed by nuclear microprobe (PIXE).
Cl (wt%)
K (wt%)
Ca (wt%)
Mn
(wt%)
Fe (wt%)
Cu (wt%)
Zn (wt%)
As (ppm)
Se (ppm)
Br (ppm)
Rb (ppm)
Sr (ppm)
Pb (ppm)
Fe/Cu
Mn/Zn
Zn/Pb
Cu/Zn
K/Ca
As/Br
Pb/Br
Mn/Br
Cu/As
1
2
3
4
5
6
7
8
54
nq
0.53
15
nq
0.78
43
nq
2.68
53
nq
0.90
7
1.6
0.90
12
2.9
1.88
24
5.0
1.51
12
2.9
0.88
25
5.4
1.3
17
5.0
1.4
0.71
nq
3.71
0.49
1216
1201
1079
2046
970
4677
0.73
nq
6.02
0.48
467
301
1454
<667
<728
3407
1.17
nq
15.44
0.43
4298
1426
900
3483
<463
1727
1.05
nq
22.2
0.47
6552
3834
800
3622
3286
2301
0.76
4.04
2.47
0.48
1155
<46
590
510
399
2055
1.35
7.38
3.61
0.82
2026
<98
1040
1773
590
3720
2.06
19.32
2.90
1.34
281
<99
1301
3264
869
7107
1.24
7.89
1.31
0.80
161
63
786
1184
427
4777
1.0
5.4
0.024
0.50
226
bdl
472
807
253
2415
0.7
4.1
0.020
0.36
276
1.64
1.59
2.33
5.16
1.77
1.96
3.48
12.9
21.4
2.05
1.65
2.20
4.41
1.56
1.95
3.58
13.0
17.8
6.66
1.54
1.88
2.17
3.33
0.22
5.46
15.8
103.3
6.00
1.55
1.67
1.65
3.33
0.20
6.08
15.7
81.6
263
2.1
2.3
0.07
5.7
0.52
5.5
23
2.1
158
0.6
1.2
0.10
4.4
0.42
3.2
9
2.1
1.46
1.04
7.59
1.53
1.41
12.5
2.73
2.47
36.1
2.23
2.04
47.3
1.13
4.33
6.59
30.5
0.32
2.34
5.04
128.9
4.78
1.92
12.9
35.9
8.19
2.88
13.1
33.9
9
10
329
713
144
1842
(1-4) - spherical brine globules co-trapped with silicate melt: (1) and (2) – shallow
and deeper globules, respectively, sample DK1-5 (Figs. 4a and 10a); (3) and (4) –
globules in coexisting silicate melt inclusions, labelled DK1-7A and DK1-7B,
respectively, in Fig. 2e (see Fig. 10b for PIXE maps of DK1-7A). (5-8) - spherical
brine globules hosted in a single quartz phenocryst, labelled DK1-8 B, C, D, E in Fig.
3 (see Fig. 10b for PIXE maps of DK1-8E). (9, 10) – average composition and
standard deviation for twenty-nine inclusions in miarolitic quartz (Kamenetsky et al.,
2002).
The concentrations were calculated by assuming a bulk density of 1.5 g/cm3, except
(3) and (4) for which a bulk density of 2.0 g/cm3 was used. PIXE analytical
parameters are given in the caption to Fig. 10.
nq – not quantified due to uncertainties caused by the presence of these elements in
the host silicate glass; bdl – below detection limit.